On Wed, 24 Sep 2014 14:02:41 -0700 Adam Williamson <adamwill@xxxxxxxxxxxxxxxxx> wrote: > On Wed, 2014-09-24 at 14:46 -0600, Kevin Fenzi wrote: > > On Wed, 24 Sep 2014 12:41:17 -0700 > > Adam Williamson <adamwill@xxxxxxxxxxxxxxxxx> wrote: > > > > ...snip... > > > > > AFAICS, the "Branch freeze" kicks in at the point we enable Bodhi > > > on the Branched tree, which is usually a couple of weeks after > > > forking it from Rawhide. For instance, on the F21 schedule - > > > https://fedoraproject.org/wiki/Releases/21/Schedule - "Branch > > > Fedora 21 from Rawhide" is listed on 2014-07-08. The first > > > 'Fedora 21 updates-testing' email I have is dated 2014-08-28, so > > > I'd say the schedule should have had an extra row, "Branch > > > Freeze", dated 2014-08-28. (The period between branching and > > > enabling Bodhi was unusually long for F21). > > > > Right. I think thats why the page was marked depreciated. > > Sorry, you think what's why? The 'deprecation' preceded the F21 cycle, > it's not related to the F21 delay. No, because in the past we used to enable bodhi at branch time. So, we would branch, and immediately have bodhi enabled. Now we don't do so until the alpha change freeze milestone. > > We have: > > > > Branch from rawhide > > branched release works like rawhide, no bodhi > > at alpha change freeze we enable bodhi and the rest of that page > > makes sense. > > Ah, OK, now I read it again, I guess the problematic part is just the > first line: > > "At the Branch Freeze event, all packages are branched from master in > source control. This is to allow the branched tree to stabilize and > enter a bug-fix and polish phase of development." > > So recently those are two separate points: the SCM branching is the > actual *branching*, when Branched comes into existence, and the > 'freeze' is when Bodhi gets turned on for the Branch. Well, that's > trivial to fix, we just re-write that line a bit. Right. > > > The name "Branch freeze" seems unfortunate to me, however, as > > > it's not really a freeze, it's more of a light cooling. I'd > > > suggest we remove the 'deprecation' notice, update any details on > > > the page which are no longer correct if anyone can see any, and > > > rename it. Ideas: > > > > > > Branch stabilization > > > Branch update policy enforcement > > > > Branch update policy seems fine to me, but really it's just the > > freeze for that milestone. So, how about: "Alpha/Beta/Final > > freeze" ? > > I still find it confusing that we have a freeze that really isn't a > freeze. I know there are different degrees of freeze and etc etc, but > it feels weird to consider the Branched tree to be in any sense > 'frozen', when it clearly isn't - outside of the 'Change Deadlines', > it gets a new package dump more or less every day. That's an odd > property for a 'frozen' tree to have, surely? > > It just feels to me like the word does not accurately describe the > actual situation, and confuses people. Oh, I see what you are getting at. Sure, Branch update policy makes sense then. For times when Branched is not in a $milestone change freeze. ...snip... > I guess I'd still find it confusing to have a "Branch Freeze" and then > Alpha, Beta and Final Freezes, but it'd certainly be an improvement on > the current situation, so if we can get consensus for the renaming of > the 'change deadlines' even without renaming the 'Branch Freeze' > that'd still be worth doing. > > Thanks for your thoughts! I'm good with Branch update policy and the milestone freezes. kevin
Attachment:
signature.asc
Description: PGP signature
-- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
